{
  "best_global_step": 10,
  "best_metric": 0.04108086973428726,
  "best_model_checkpoint": null,
  "epoch": 2.96,
  "eval_steps": 5,
  "global_step": 36,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.4,
      "grad_norm": 0.029360348358750343,
      "learning_rate": 8.888888888888888e-07,
      "loss": 0.0621,
      "step": 5
    },
    {
      "epoch": 0.4,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 803.975,
      "eval_kl": 3.279149532318115e-05,
      "eval_loss": 0.050603967159986496,
      "eval_num_tokens": 73036.0,
      "eval_reward": 0.075,
      "eval_reward_std": 0.12886751294136048,
      "eval_rewards/equation_reward_func": 0.0625,
      "eval_rewards/format_reward_func": 0.0125,
      "eval_runtime": 1333.6061,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 5
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.01968163251876831,
      "learning_rate": 7.5e-07,
      "loss": 0.0365,
      "step": 10
    },
    {
      "epoch": 0.8,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 783.9375,
      "eval_kl": 3.088414669036865e-05,
      "eval_loss": 0.04108086973428726,
      "eval_num_tokens": 146337.0,
      "eval_reward": 0.0875,
      "eval_reward_std": 0.15386751294136047,
      "eval_rewards/equation_reward_func": 0.0875,
      "eval_rewards/format_reward_func": 0.0,
      "eval_runtime": 1299.4728,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 10
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.0,
      "learning_rate": 6.111111111111112e-07,
      "loss": 0.0461,
      "step": 15
    },
    {
      "epoch": 1.24,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 803.8875,
      "eval_kl": 3.0159950256347656e-05,
      "eval_loss": 0.08998335152864456,
      "eval_num_tokens": 219977.0,
      "eval_reward": 0.075,
      "eval_reward_std": 0.12886751294136048,
      "eval_rewards/equation_reward_func": 0.05,
      "eval_rewards/format_reward_func": 0.025,
      "eval_runtime": 1335.4186,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 15
    },
    {
      "epoch": 1.6400000000000001,
      "grad_norm": 0.010507287457585335,
      "learning_rate": 4.722222222222222e-07,
      "loss": 0.0786,
      "step": 20
    },
    {
      "epoch": 1.6400000000000001,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 859.9375,
      "eval_kl": 3.051459789276123e-05,
      "eval_loss": 0.07127637416124344,
      "eval_num_tokens": 298506.0,
      "eval_reward": 0.15,
      "eval_reward_std": 0.21969234347343444,
      "eval_rewards/equation_reward_func": 0.1,
      "eval_rewards/format_reward_func": 0.05,
      "eval_runtime": 1288.7661,
      "eval_samples_per_second": 0.016,
      "eval_steps_per_second": 0.004,
      "step": 20
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.008569754660129547,
      "learning_rate": 3.333333333333333e-07,
      "loss": 0.0836,
      "step": 25
    },
    {
      "epoch": 2.08,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 794.325,
      "eval_kl": 3.255009651184082e-05,
      "eval_loss": 0.05673029273748398,
      "eval_num_tokens": 377844.0,
      "eval_reward": 0.0625,
      "eval_reward_std": 0.10386751294136047,
      "eval_rewards/equation_reward_func": 0.0625,
      "eval_rewards/format_reward_func": 0.0,
      "eval_runtime": 1326.7789,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 25
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.014444947242736816,
      "learning_rate": 1.9444444444444445e-07,
      "loss": 0.0494,
      "step": 30
    },
    {
      "epoch": 2.48,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 750.6,
      "eval_kl": 2.9385089874267578e-05,
      "eval_loss": 0.12568210065364838,
      "eval_num_tokens": 454764.0,
      "eval_reward": 0.15,
      "eval_reward_std": 0.23660253882408142,
      "eval_rewards/equation_reward_func": 0.125,
      "eval_rewards/format_reward_func": 0.025,
      "eval_runtime": 1334.1012,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 30
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.013897239230573177,
      "learning_rate": 5.555555555555555e-08,
      "loss": 0.0519,
      "step": 35
    },
    {
      "epoch": 2.88,
      "eval_clip_ratio": 0.0,
      "eval_completion_length": 794.9,
      "eval_kl": 3.1581521034240724e-05,
      "eval_loss": 0.048388056457042694,
      "eval_num_tokens": 525128.0,
      "eval_reward": 0.075,
      "eval_reward_std": 0.10773502588272095,
      "eval_rewards/equation_reward_func": 0.05,
      "eval_rewards/format_reward_func": 0.025,
      "eval_runtime": 1313.9163,
      "eval_samples_per_second": 0.015,
      "eval_steps_per_second": 0.004,
      "step": 35
    }
  ],
  "logging_steps": 5,
  "max_steps": 36,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 15,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}